Artificial bee colony-based extraction of non-taxonomic relation between symptom and syndrome in TCM records (Q2224057)

Summary: Research the extraction of non-taxonomic relation between symptom and syndrome in Traditional Chinese Medicine (TCM) records. In the present study, a niche-based artificial bee colony (ABC) algorithm is proposed to address a variety of problems (e.g., low efficiency, slow convergence rate, miss-reporting and etc.) of conventional rule-oriented extraction of non-taxonomic relation. By combing the vast merging/evolution diversity of niche technique and rapid non-taxonomic relation extraction of ABC algorithm, the proposed algorithm was able to resolve the problems of local optimum and rules redundancy. In addition, confirmatory experiment is performed on TCM medical record corpus; the results show that, when compared with the conventional relationship rule mining algorithm, the novel algorithm featured significant improvements in both the individual diversity and the efficiency of extracting effective rules, and thus excavating the results which can be used as references for extraction of non-taxonomic relation between symptom and syndrome in TCM medical records.